MIPT Researchers Train Neural Network to Predict Rock Permeability
It runs faster than conventional tools while delivering predictions of comparable accuracy.
Researchers at the Moscow Institute of Physics and Technology (MIPT), working with colleagues from Iran, have developed a neural network algorithm that predicts rock permeability based on structural characteristics.
Such calculations are critically important to the extractive industries. Permeability determines how efficiently a deposit can be developed. It affects how much oil or gas can be recovered from a reservoir and how quickly contaminants can spread through groundwater.
The new technology can fully replace laboratory studies and supercomputer-based computations. To train the neural network, the researchers used a dataset of 998 synthetic three-dimensional images of porous media. Rather than using the 3D images themselves as input, they used compact statistical descriptors known as correlation functions.
